Sect_9.1.1

9.1.1 Introduction to GPS Network Processing

INTRODUCTION

In general, a GPS survey campaign involves the use of a small number of receivers to coordinate a large number of stations. The area of survey operations may span distances of merely a few kilometres (as on an engineering site), to several hundred kilometres, or even thousands of kilometres in the case of geodynamical surveys. A typical GPS survey, such as for mapping or control densification, involves distances of the order of several tens of kilometres. The survey may be carried out using conventional static GPS survey techniques, or the modern "high productivity" techniques described in section 5.5.1. However, the principles of network processing are the same whether a small number of baselines were observed over several days using conventional GPS, or many baselines observed in a matter of a few hours.

In a GPS network survey a number of processing strategies are possible:

If the number of receivers that have been deployed during an observation session is greater than two, then the appropriate single session processing strategy must take into account multiple baselines.
As the survey cannot be completed during a single session of observations, a suitable , involving the multi-session processing strategy propagation of the results of one session solution into another (and eventually across the entire network) has to be used.

Network Processing Terminology

A GPS "network solution" is a generic term for any GPS solution involving two or more stations. It may be: (a) as small as a subset of one session (for example, a single baseline), (b) an entire session, or (c) a number of sessions. It may be derived from a reduction of the phase data itself (as in the baseline processing described in Chapter 7 and Chapter 8), or computed from a secondary adjustment of GPS baseline results.
A GPS "campaign solution" involves all the stations within a survey network that: (a) have been coordinated in a single field operation (usually involving a number of observation sessions) and identifiable as a single "job" for a client, and (b) which are adjusted together, transformed and perhaps integrated into an existing geodetic network.
It is possible to distinguish between directly-connected stations (those observed in the same session, and adjusted together in a single session solution), and the indirectly connected stations whose relative coordinates are derived from a multi-session solution.
Both single and multi-session network solutions may be obtained from either: (a) a primary adjustment of the raw GPS phase data (if the appropriate phase reduction software is available), or (b) a secondary adjustment of the GPS baseline results (themselves the output of a primary phase adjustment).
It is possible to distinguish between a minimally constrained network solution in which only one station (the so-called "datum" station) has been held fixed, and a GPS network solution that has been constrained to fit an existing geodetic network.

Single Session Processing

The following are the single session processing strategies:

The SINGLE BASELINE processing mode. In this case the primary GPS reduction software can only handle single baselines. The individual baselines are processed one by one, and the output coordinates and variance-covariance (VCV) matrices are input into the secondary network adjustment software. The correlations between the baselines and between the differenced data are neglected.

The MULTI-BASELINE processing mode. This is the mathematically rigorous mode of single session phase data processing because the correlations between the baselines are taken into account. However, in forming the double-differences only the independent baselines are used. This mode of processing therefore takes away the "arbitrariness" of single session processing.

There is therefore an increase in mathematical rigour for session adjustments, from the single baseline mode to the multi-station mode. Furthermore, ambiguity resolution may be easier in the context of multi-baseline and multi-station session processing than if the baselines are processed independently. This is because the correlations in double-differenced observables between baselines aid the resolution procedure.

Multi-Session Processing

In relation to GPS surveying involving more than one session, the following points should be noted:

The GPS data collected during one session has special properties. The dataset for a particular GPS session is independent of any other data collected during any other session.

It may therefore be processed separately.
However, if there are stations that are common to two (or more) observation sessions, the station coordinates, or baselines, are (functionally) correlated and the coordinate results of one session solution will influence the results of another solution.

If there is only one common station between sessions, then each session solution may be considered to represent a separate minimally constrained GPS solution.
To combine session solutions, a minimum of one station must be common to two sessions in order to provide the connection between sessions, and to maintain the "minimally constrained" nature of the GPS network as it is built up session by session. More than the minimum number of connections increases the redundancies in the survey (see Figure below).

Moving from session to session: multiple station occupancy.

This has two effects: (a) it improves the overall quality and reliability of the network, and (b) it means that the most rigorous form of adjustment must be a simultaneous multi-session reduction of GPS phase data.

The manner in which each session is linked to the previous session (and hence back to the original Datum Station) and to the next: (a) is an important consideration during the network design stage, and (b) has implications for the processing strategies for the total (multi-session) dataset.

The combination of separate session solutions can be carried out in a number of ways:

Using multi-session GPS phase data reduction software. This ensures that stations that are common to more than one session are correctly weighted in the solution, and that datum transfer is carried out within the matrix operations of the Least Squares adjustment. High precision geodetic (or "scientific") GPS software all have multi-session processing capability.
Secondary network adjustment software can use the results of individual session solutions as input. The multi-session adjustment results and VCV matrices reflect the (independent) contributions of the sessions. The matrices are altered to maintain the minimally constrained nature of the total adjustment. GEOLAB^TM is a well-known network adjustment program which can accept GPS coordinates and baselines as observations, as well as traditional survey measurements such as slope distances, theodolite directions, etc.
The task can also be performed with specially written software to handle only the GPS output (that is, no conventional distance and direction data to be used, unlike the case of the GEOLAB^TM program). The software is relatively simple as it merely concatenates the individual session VCV matrices.

The most rigorous processing of multi-session data, collected in a field campaign in which redundant station occupations were made, is therefore the simultaneous reduction of all phase data in one step.

Back To Chapter 9 Contents / Next Topic